AI agency is a TRAP - I'm shutting it down

Key Concepts

• AI Agency Model (Traditional): Building custom AI solutions for clients, often involving significant manual effort and limited scalability.
• AI Agent: Autonomous AI systems designed to perform specific tasks, often involving interaction with various tools and APIs.
• Outcome-Based Pricing: Charging clients based on results achieved (e.g., qualified appointments booked) rather than time spent or services rendered.
• Productized System: A repeatable, scalable process automated with AI to deliver consistent results.
• ICP (Ideal Customer Profile): A detailed description of the perfect customer for a business.
• LLM (Large Language Model): A type of AI model, like GPT-3 or GPT-4, capable of understanding and generating human-like text.
• Data Enrichment: The process of enhancing data with additional information from various sources.
• Voice Cloning: Creating a digital replica of a person's voice using AI.

The Broken AI Agency Model & Shift to Outcome-Based Systems

The speaker details his experience building an AI agency that reached $80,000 in monthly revenue but ultimately proved unsustainable, leading to its closure. He argues the traditional AI agency model is fundamentally broken due to two key reasons: lack of scalability and increasing commoditization of AI services.

Financial Realities of the Traditional Model: Despite $80,000 in revenue in their best month, profit margins were significantly lower. Approximately $56,000 went to contractors, $3,000 to software, and at least $12,000 to other business expenses, leaving less than 30% as profit. This starkly contrasts with the inflated income claims often seen online. As the speaker states, “80K a month on Stripe does not nearly mean 80K in your pocket.”

Reason 1: Lack of Scalability – The Customization Problem: Unlike traditional agencies (SMMA, web design) with standardized outputs, AI projects are inherently custom. Even seemingly similar requests evolve into unique projects with varying requirements, integrations, and deployment needs. He cites a client initially seeking an AI rack system who ultimately required a full software product with custom UI, backend, and deployments. This leads to the agency functioning more as a development shop or project management service, hindering scalability and IP development. Scaling increases complexity, not efficiency. The speaker notes that only those with large platforms (300k+ YouTube subscribers) can charge premium rates to offset these costs.

Reason 2: Commoditization – AI Gets Better, Demand for Agencies Decreases: As AI tools become more accessible and powerful (Cursor, Cloud Code), the need for custom AI solutions diminishes. Tasks previously requiring developers can now be handled directly by business users with prompting skills. The speaker highlights that even internal company knowledge chatbots, once $3-5k projects, can now be set up in 30 minutes. He observes a trend towards “AI transformation partner” models (consulting-focused) but predicts even this will be short-lived as AI becomes a core skill for knowledge workers. He emphasizes that “this business model gets worse as AI gets better and most likely it won't even exist by the end of 2026.”

The New Approach: Appointify – Outcome-Based, AI-Powered Lead Generation

The speaker transitioned from the traditional agency model to a new system called Appointify, focused on delivering qualified appointments booked directly onto clients’ calendars, with payment only for confirmed appointments. This shift dramatically improved margins to over 70%.

Core Offering: Qualified appointment booking with a replacement guarantee if a lead is unqualified. The pricing is $250 per qualified appointment.

AI-Powered Automation: The entire process is heavily automated using AI agents and various tools:

• ICP & Offer Document Generation: AI agents create Ideal Customer Profile (ICP) and compelling offer documents.
• Mailbox Setup & Warm-up: Automated mailbox setup and warm-up using AI agents.
• Lead Scraping: AI agents generate LinkedIn searches based on the ICP and scrape leads from platforms like LinkedIn Sales Navigator, Apollo, and Google Maps.
• Data Enrichment: Leads are enriched with data from Clay, including company details, personal information, and LinkedIn posts.
• Email Finder: Email addresses are found using an email finder action within Clay.
• Voice Cloning: Client voices are cloned using 11Labs for personalized outreach.
• Loom Video Personalization: AI generates personalized Loom videos featuring the prospect’s name and website, significantly improving conversion rates.
• Campaign Management & Optimization: An agent tracks campaign performance in a Google Sheet, adjusting parameters based on KPIs.

Cost Breakdown: After onboarding, a booked appointment costs approximately $11 in API and infrastructure costs, plus subscription fees for tools like Clay and Instantly.

Key Result: Two clients onboarded, generating four appointments for one client in a week, resulting in $1,000 in revenue.

The Framework for Building a Scalable AI-Powered Business (2026 Roadmap)

The speaker outlines a step-by-step roadmap for building a successful AI-powered business in 2026, emphasizing the importance of focusing on outcomes rather than just AI services.

Week 1: Niche Selection: Choose a service-based industry with existing revenue (e.g., accounting, SEO). Research the industry and its processes.

Week 2: Manual Validation: Sell the outcome manually, positioning yourself as a traditional agency. Focus on delivering value without relying on AI initially. He cautions against branding heavily with AI at this stage.

Week 3: System Building (Repetitive Tasks): Identify repetitive tasks within the process and build an AI system to automate them. He specifically mentions using GPT-4 for tasks like CSV ingestion, categorization, and P&L statement generation.

Week 4: Iteration & Refinement: Gather feedback from real users and iterate on the system until the output is consistent and reliable.

Long-Term Vision: Create a productized system that can be scaled repeatedly with minimal manual effort.

Notable Quotes

• “80K a month on Stripe does not nearly mean 80K in your pocket.”
• “This business model gets worse as AI gets better and most likely it won't even exist by the end of 2026.”
• “Your goal should not be to sell AI agency services. Today, your main goal should be to build AI systems.”
• “Your main goal should be to build AI systems…business owners no longer want to experiment. They want proven workflows that bring consistent results.”

Synthesis & Conclusion

The speaker’s experience provides a cautionary tale about the pitfalls of the traditional AI agency model. He convincingly argues that the model is unsustainable due to its lack of scalability and the rapid commoditization of AI services. His transition to Appointify demonstrates a more viable path forward: focusing on delivering measurable outcomes, leveraging AI to automate the delivery process, and pricing based on value provided. The roadmap he presents offers a practical framework for building a scalable, AI-powered business in 2026, emphasizing the importance of identifying and automating valuable processes rather than simply offering “AI services.” The core takeaway is a shift in mindset: from selling what you do (AI solutions) to selling the results you achieve (qualified appointments, increased revenue, time savings).